Download AN INTERPRETATION OF GLASS CHEMISTRY IN TERMS OF THE

Journal of Non-Crystalline Solids 21 (1976) 373-410
© North-Holland Publishing Company
AN INTERPRETATION OF GLASS CHEMISTRY IN TERMS OF THE
OPTICAL BASICITY CONCEPT
J.A. DUFFY and M.D. INGRAM
Department of Chemistry, University of Aberdeen, Meston Walk, Old Aberdeen,
Aberdeen AB9 2Utf, Scotland, UK
Received 17 June 1975
Revised manuscript received 2 October 1975
The basicity of an oxide glass can be measured experimentally from the frequency
shifts in the ultra-violet (UV) (s-p) spectra of probe ions such as Pb2+ and can be expressed on the numerical scale of optical basicity A (ideally A lies between zero and
unity). It is possible to relate A with (i) the constitution, and (ii) the electronegativity
of the cations (e.g. Na+, Si4+, etc.) of the glass, and the relationship allows microscopic
optical basicities ~ to be assigned to individual oxides and oxy-groups in the glass. These
microscopic optical basicities are used for interpreting various aspects of the physics and
chemistry of glass including refractivity, network coordination number changes, chemical
durability, the glass electrode, UV transparency and the host behaviour of glass towards
metal ions generally. Changes in glass basicity in going from one alkali metal oxide to
another are also discussed. Finally, the concept of optical basicity, both as an experimentally obtained quantity and as a number calculated from glass constitution and electronegativity, is discussed in relation to the traditional approach to acid-base behaviour
in glass.
1. Introduction
Commercially important glasses consist essentially of metal silicates, often with
smaller quantities of phosphate, borate and other anions, and through the ability of
the silicon, phosphorus and boron to form bridges with oxygen, these oxyanions
exist largely as chains or networks. The formation of oxyanion glasses can be envisaged
in terms of the reaction between a metal oxide (e.g. Na20, CaO, etc.) and an acidic
oxide (e.g. SiO2, B203, etc.), and because of the wide range of compositions attainable, products of varying degree of acidity and basicity can be obtained. The basic
nature of the oxygen atoms is an important feature and has a profound bearing on
the physics and chemistry of the glass. The measurement of basicity in a glass is not
straightforward. Indeed, the term "basicity" is open to a number of interpretations,
but in the context of this paper it refers to the "state" of the oxygen atoms and how
they would react, for example, to the acid action of solute metal ions. According to
this viewpoint, addition of a small concentration of a solute metal ion to a glass resuits in the oxygens of the glass donating some of their negative charge to the metal
373
374
J.A. DullY. M.D. Ingrain/An interpretation of glass chemistry
ion. The oxygens behave as bases in the Lewis sense and correspondingly the metal
ions behave as Lewis acids.
The ability of oxygen to donate negative charge is at a maximum when it exists
as the "free" 0 2- ion uninfluenced by surrounding cations; this situation is approached when the cations are almost non-polarizing, for example Na ÷, K ÷ or Ca 2+.
When oxygen is attached to atoms such as silicon, as in terminal S i - O or bridging
S i - O - S i units, its basicity is much less. A convenient way of regarding this lowering
of the basicity is to envisage the oxygens in the S i - O or S i - O - S i units still as 0 2ions but influenced by their interaction with the highly polarizing Si 4÷ ion. In the
terminal S i - O unit it is influenced by one Si 4+ ion, while in the S i - O - S i unit it is
influenced by two. The polarization of 0 2- results in negative charge being drawn
off the ion (i.e. covalent bond formation), and the oxygen is therefore less able to
donate charge to a solute metal ion, i.e. it is less able to function as a Lewis base.
The overall basicity of a glass containing both bridging and non-bridging oxides will
depend upon the relative proportion of these two types of oxygen, and this will also
determine the total electron donation to the solute metal ion.
Certain metal ions undergo observable changes (e.g. a colour change or a change
of oxidation state) depending upon the degree of electron donation they receive
from the oxygens, and therefore may be used as "probes" for basicity in glasses.
One of the earliest studies of glass basicity using a metal probe ion was by Weyl and
Thfimen [1] in 1933 who exploited the Cr(III)/Cr(VI) equilibrium. Other examples
of metal ion probes are the cobalt(II) ion (which undergoes an octahedral-tetrahedral change in stereochemistry at a critical basicity) and chromium(VI) (which
changes from a dichromate species to CRO42-) -- see below.
All of these indicator ions suffer from the disadvantage that their use is necessar1"O
II
=
/
\
/
o
,
\
j
I
t
s
|
~,000
i
i
40,000
(ore-,)
30,000
Fig. 1. (a) Absorption spectrum of Pb 2+ in sodium borate (20% Na20) glass ([Pb2+] = 0.00113 M,
path length 1.50 ram); spectrum (b) of Pb 2+ in 11 M HC1 is given for comparison.
J.A. Dully, M.D. lngram / An interpretation of glass chemistry
375
ily limited to the one narrow region of basicity at which the critical change in the
metal ion occurs. Their action is analogous to the colour change of, for example,
methyl orange or litmus in aqueous acid-base equilibria. Furthermore, the way in
which they signify a change in basicity is indirect and often depends upon specific
conditions such as the 0 2 partial pressure of the atmosphere in which the glass was
melted. A more satisfactory type of probe ion is one which responds in a direct
and, more importantly, a gradual manner to changes in glass basicity. Recently, it
has become apparent that probe ions which respond in this way are p-block metal
ions in oxidation states two units less than the number of the group to which they
belong, e.g. TI+ (group Ili), Pb 2÷ (group IV), Bi3÷ (group V). The electronic configurations of these metal ions is such that there is a pair of electrons in the outermost (6s) orbital, and when small concentrations of the ions are dissolved in glass,
the 6s -+ 6p electronic transition gives rise to an intensely absorbing ultraviolet (UV)
band, usually with a sharp maximum. A typical absorption spectrum is shown in
fig. 1 for the Pb 2+ ion in a borate glass and it can be compared with the corresponding spectrum for the chloro complex of lead(ll) in hydrochloric acid. The absorption band is rather more broad in the glass, but in view of the enormous frequency
shifts which changes in basicity can produce in these p-block metal ions, it is apparent that the probe ions tend to select closely similar sites rather than seek out
a wide range of different sites *. Spectral measurements [4-7] of T1+, Pb'2+ and
Bi 3+ in glasses have shown that the frequency of the absorption band is dramatically lowered by increased basicity of the glass, and the reason for this appears to be
due to orbital expansion effects within the probe ion brought about by electron
donation by the oxygens. This effect, known as the "nephelauxetic" effect, is well
known in inorganic spectroscopy and is understood in sophisticated terms largely
as a result of the pioneering studies of JCrgensen [8] during the 1950s and 1960s.
JOrgensen's studies were concerned essentially with transition metal ions, but it
has been shown that the nephelauxetic effect for p-block metal ions is fundamentally linked to that for transition metal ions [9,10].
To understand the spectroscopic shifts of p-block metal ions in their response
to basicity and to appreciate how the orbital expansion comes about, it is necessary
to view the interaction between the p-block metal ion and surrounding oxygens in
terms of the molecular orbital theory. This has been discussed by the present authors
in ref. [11]. (Incidentally, it is important to be aware that arguments based upon
valence bond theory and orbital hybridization are inappropriate.) However, in simple
terms, the decrease in frequency of the 6s ~ 6p transition can be thought of as a
consequence of the covalency effect in which the electrons received from the oxygens are accommodated in o and rr bonding molecular orbitals [ 11 ]; some of this electron density is located between the inner electron core of the metal ion and its 6s
orbital. This is shown schematically for the Pb 2+ ion in fig. 2. One of the factors affecting the energy of the 6s ~ 6p process is the force of attraction that the 6s electron exper* The converse sometimes occurs, as for instance in bromide-sulphate glasses [2] or in alkali
borate glasses at elevated temperatures [3].
376
J.A. DullY, M.D. Ingram / An interpretation o f glass chemistry
nucleus
I
I
~
/
~
I
I
inner
/
~
i
N
~
~
I
~
~
electron
/
6
s
.
density
orbit /
,
]
/
~ / - ' " J
extra electron density
(donated by oxides~
(a)
(b)
Fig. 2. Schematic diagram of (a) free Pb 2+ ion, and (b) Pb 2+ ion after receiving negative charge
from neighbouring oxide ions.
iences from the nucleus. Much of the positive pull of the nucleus is screened by the inner
electron core, but the electron density donated by the oxygens serves to increase this
screening further and this in turn allows the 6s electron to escape more easily to the
6p level (even though the 6p level is also screened). Electron donation by the oxygens
to the probe ion therefore brings about a reduction in the 6s - 6p energy difference,
and therefore a reduction in frequency in the UV absorption band, compared with
the free Pb 2+ ion.
The question now arises as to what is meant by the free Pb 2+ ion shown in fig. 2.
The free Pb 2+ ion is in fact only a hypothetical entity, but its properties may be calculated by extrapolation of the linear relationship between measured spectroscopic
shifts and orbital expansion parameters for various ligand environments [9]. It is assumed that when the orbital expansion parameters of the ligands are extrapolated to
zero, then the "electron donating ability" of the ligands is zero also; in effect they
are able to provide an "unperturbing" environment for free Pb 2+ ions. The extrapolated value of the 1S0 ~ 3P 1 transition (which represents the energy difference
corresponding to the 6s -+ 6p process) for Pb 2+ is found to be 60 700 c m - 1, and
this will be referred to subsequently as Vfreeion [9]. It is worth noting that it is
close to the experimentally determined value (64 390 cm -1) for Pb 2+ ions observed
in gas discharges.
1.1. Optical basicity
The spectroscopic shifts observed in the 1S0 ~ 3P 1 bands of T1+, Pb 2+ and Bi 3+
have been used by us for setting up scales of basicity [5]. However, most data have
been obtained for the Pb 2+ ion and we shall centre our attention upon this ion. For
pb2+,/"free ion is 60 700 cm-1 (see above) and Uo2- , the frequency of Pb 2+ in an
J.A. Duffy, M.D. Ingrain/An interpretation o/glass chemistry
377
"ionic" oxide, is 29 700 cm -1. Thus on going from the unperturbed Pb 2+ ion to the
situation where the Pb 2+ ion receives the electron density from free oxides, there is
an enormous spectroscopic shift of 31 000 cm-1. The frequency of 29 700 cm-1 is
that for Pb 2+ in CaO *, and no doubt if some other metal oxide, e.g. Na20 or K20,
were chosen for providing an environment of free oxides for Pb 2÷ a slightly different (and lower) frequency would be obtained, leading to a shift of even greater than
31 000 cm -1. However, since an environment containing ideally ionic oxide ions
certainly does not exist, the choice of metal oxide is necessarily arbitrary (see fig. 3).
When the probe ion is influenced by the less basic oxides of a glass, the frequency
shift is somewhat less. For example in C a O - P 2 0 5 (1 : 1) glass the frequency of
Pb 2÷ is 46 200 cm -1 while in N a 2 0 - S i O 2 (3 : 7) glass it is 42 200 cm -1, corresponding to shifts of 14 500 and 18 500 cm -1, resepctively. The spectroscopic shift represents the electron donor power of the oxides in the glass, and for comparing oxide
glasses with each other it is convenient to express the basicity as the ratio of the
electron donor power of oxides in the glass to the electron donor power of free
oxide ions. This ratio is derived entirely from spectroscopic measurements and accordingly the glass basicity has been termed [5] "optical" basicity A. Using Pb 2+ as
the probe ion:
Pfree ion - l)glass AVglass At'glass
(1)
Apb(II) = Pfree ion -- VO2 - - AI)O2-- 31 000 '
where Vglass is the frequency of the Pb 2+ ion in the glass under investigation. By definition, Apb(ll)(CaO ) is unity.
In terms of specific examples, spectroscopic shifts (AI)glass) yield optical basicities,
Apb(iI) as follows: CaO - P205 (1 : I) glass, AVglass = (60 700 - 46 200) and
Apb(ii) = 0.47; Na20 - SiO 2 (3 : 7) glass, AUglass = (60 700 - 42 200) and Apb(ii) =
0.60.
Although most optical basicity data have been obtained using Pb 2+ as a probe
ion, T1+ and Bi 3+ have also been used, and the agreement between the values obtained using the three ions has been found to be good [5,12]. With these ions, the
optical basicities are calculated as follows [5] :
ATI(I) = ( 5 5 300 - l)glass)/18 3 0 0 ,
(2)
ABi(III ) = (56 000 - Vglass)/28 8 0 0 .
(3)
1.2. Experimental
In the practical determination of optical basicity, the fragments of glass for
spectroscopic examination need be only a few millimetres in cross-section; they are
* The introduction of Pb 2+ ions into an ionic metal oxide such as CaO may at first sight be expected to produce a lead-oxygen interaction which is predominantly ionic. However, such an
expectation does not take account of the enormous polarizability of which the so-called "hard"
oxide ion is capable.
J.A. DullY, M.D. lngram / An interpretation of glass chemistry
378
mounted over a 3 - 4 mm dia. aperture of a mask that is placed in the light beam of
the spectrophotometer (the reference beam being similarly attenuated). It is best to
cast the glass with a path length of ~ l m m (for which a probe ion concentration of
10-3M is appropriate) to keep the absorbance of the glass as low as possible. If
the glass itself absorbs significantly in the region of the probe ion absorption band,
then it is necessary to correct the absorption using the spectrum of the undoped
glass (of the same path length). The frequency of the absorption maximum can. usually be determined to within +-50 cm-1 (or sometimes better) and this corresponds
to -+0.001 unit of optical basicity. For obtaining the optical basicity of a number
of glasses, the "platinum loop" technique of Easteal and Udy [7] is very useful.
At higher frequency than the 1S0 ~ 3P 1 band is also the IS 0 ~ 1P 1 band. The
latter band is approximately five times more intense than the former and occasionally slight overlapping results in the requirement for a correction in frequency of
the 1S0 ~ 3P 1 band; this correction is usually quite small ( ~ 100 cm-1).
1.3. Theoretical prediction o f basicity values
Let us develop further the idea that the oxygen atoms in, say, CaO-P205 (I : 1)
glass are oxide ions influenced very strongly by the p5+ cations and less strongly
by the Ca 2+ cations in the glass. In this particular glass, the Ca 2+ ions neutralize
one-sixth of the negative charge of the oxide while the p5+ ions neutralize five-sixths.
The lowering of the electron donor power of the oxides by the Ca 2÷ and p5+ ions
[which corresponds to a frequency difference of (46 2 0 0 - 2 9 700) cm -1, fig. 3]
might be expected to be due to (i) the proportion of negative charge each cation
neutralizes, and (ii) some polarizing or electron-attracting property of each cation.
This latter property can be evaluated from experimental spectroscopic shifts and is
known as the "basicity moderating power" 3' [ 13, 14]. This parameter appears to
have fundamental significance, and indeed has been shown to be related to the
Pauling electronegativity x by the equation 3' = 1.36 (x - 0.26).
E
? .....
ideally
ionic
0 2-
02- ions
in CaO
02- ions in
CaO: P20s (1:1)glass
('Vo2- =
('1/glass =
29, 7°°_ _e~'_)__..4_6_,20_9__em_"_). . . . . . . . . .
30,000
{.Vglass-~'02- )
20,000
"Vfreeion-'V02-}
O"
P
A
J0
Q.
10,000
M .... i /freeion-'Vglasl
Decreasing basicity
•
"Vfree ion
( 6 0 , 7 0 0 cm"l )
Fig. 3. Pictorial representation of electron donor power (height of shaded column = A~,)of oxide
under various conditions. The frequencies are those registered by Pb2+ probe ions in sensing
these basicities.
J.A. Duff.v, M.D. Ingrain / A n interpretation o f glass chemistry
379
From the manner in which the basicity moderating parameters have been evaluated,
it is possible to express the contribution to the lowering of the electron donor power
of oxide in C a O - P 2 0 5 (1 : 1) glass by
(a) calcium ionsas (Vfree ion - v 0 2 - ) " ~ (1 - l/Yea ) ;
(b) phosphorus ions as (Vfree ion - v02- ) " s (1 - 1/Tp) •
Thus the total effect produced by both cations corresponds to the frequency difference of
(/)free ion -- PO2 - ) (-~(1 -- 1/3'Ca) + ~(1 -- 1/3'p)},
experimentally this is equal to (46 200 - 29 700) cm -1. Since, as discussed above,
the oxide ions in CaO are regarded as free oxide ions with the optical basicity of
CaO as unity, 7Ca must be unity; on the other hand the cation p5+ is expected to
be highly polarizing and so 7p is expected to be considerably greater than unity.
(The value of 7p is in fact 2.50; "r-values for other elements are in table 1.)
For oxide glasses generally, the lowering of the electron donor power o f the
oxides, expressed in terms of the frequency shift in going from CaO to the glass is
given by
(/;glass -- VO2-) = (/)free ion - VO2-){(ZArA/2)(1 -- 1/~'A) + (zB%/2)(1 -- 1/~'B) + --'},
where z A, z B ... are the oxidation numbers of the cations A, B ... and r A, r B... are their
ionic ratios with respect to the total number of oxides. Dividing the equation by
(/2free ion - / 2 0 2 - ) and replacing (/2glass - / 2 0 2 - ) by ((/2free ion - / 2 0 2 - ) - (/2free ion Table 1
Basicity moderating parameters, %
Element
"7
chlorine
nitrogen
sulphur
carbon
phosphorus
hydrogen
boron
silicon
zinc
aluminum
magnesium
calcium
lithium
sodium
potassium
rubidium
caesium
3.73
3.73
3.04
3.04
2.50
2.50
2.36
2.09
1.82
1.65
1.28
1.00
1.00
0.87
0.73
0.73
0.60
380
J.A. DullY, M.D. Ingram / A n interpretation o f glass c h e m i s t r y
Table 2
Formulae for evaluating ideal optical basicity of borate, silicate and phosphate glasses containing
x% of alkali oxide (M20).
Glass
A a)
borate (xM20: (100-- x)B203
x
3(100 - x)
7M(300 - 2x) + 2.36 (300 - 2x)
silicate (xM20:(100 - x)SiO 2)
x
2(100 - x)
7M(200 - x) ÷ 2.09(200 - x)
phosphate (xM20:(100 - x)P20 s
__
x
5(100 - x)
3,M(500 - 4x) + 2.50(500 - 4x)
a) Basicity moderating parameters ")'Mfor the alkali metals are given in table 1.
Uglass)) (see fig. 3) yields on rearrangement *
A = 1 - ((ZArA/2)(l
-- 1/TA) + ( Z B r B / 2 ) ( 1
-- 1/')'B) + ...).
(4)
This formula allows A to be calculated for any oxide glass f r o m its chemical constitution and from the basicity m o d e r a t i n g parameters o f the various cations, e.g. Na +,
Si 4+, B 3+, etc., present in the glass and good agreement is f o u n d w i t h e x p e r i m e n t a l
A values [13, 14].
The value o f A o b t a i n e d theoretically is an " i d e a l " optical basicity which is the
0"6
,0
[7
.0""
sJ~ s
o.s
0"4
I
I
I
I
I
0
10
20
30
40
% Na,O
Fig. 4. Trend in ideal optical basicity in the Na20-B203 glass system (continuous line). Experimentally determined optical basicities are denoted: [] TI*, ® Pb 2+ and/x Bi3+.
* Equation (4) is presented rather than the simpler form [ 14]
A = (ZArA/2"r A) + (ZBrB/27B) + ...
since the former is more versatile and enables the calculation of, for example, microscopic
optical basicity values (see below).
J.A. DullY, M.D. Ingram / An interpretation of glass chemistry
381
average of all the oxides in the glass. It is possible to plot the trend in ideal optical
basicity as the alkali content of a glass increases, and the appropriate formulae for
borate, silicate and phosphate glasses are given in table 2 *. The trend for the
N a 2 0 - B 2 0 3 glass system is shown in fig. 4, together with experimental values of
ATI(1), Apb(ii) and ABi(III). Up to about 16% Na20, agreement between experimental and ideal basicities is good, but with increasing Na20 content there is increasing
disagreement between the different experimental values. This unexpected effect has
been discussed at length elsewhere [ 12,15,16]. We have suggested that the answer
lies in the range of basicity values which is needed to characterize the sites present in
a partially disrupted network glass, but phase separation may also be involved. It is
interesting to note that ABi(III) reflects most accurately the trend in ideal basicity.
1.4. Microscopic optical basicity, X
As well as providing a means of calculating A for a bulk glass, eq. (4) makes it
possible to calculate the microscopic optical basicity of individual oxygen atoms in
any oxyanion unit.
Let us consider for example the oxygen atoms in the SiO 4 - anion. We imagine
first of all that there are no polarizing cations present other than the Si4+ ions, i.e.
the other metal ions present are alkali or alkaline earth ions. Thus since there are
four terminal oxides attached to each Si 4+, the Si: O ratio rsi = ~; therefore
zsirsi/2 = (4 X ~)/2 =1~. To distinguish microscopic basicity from bulk basicity, we
replace A by X, and eq. (4) yields X = 1 ½(1 - 1/7). With Tsi = 2.09 (see table 1),
X = 0.74.
If we now consider an oxide which, instead of being terminal, serves to bridge
two silicons, the polarization effect on the oxide will be twice as great since it is being affected by two Si 4+ ions. Eq. (4) now yields
X=l-2X~(1-
1/7Si)
and X is therefore 0.48.
Now let us consider an oxide that bridges, say, a silicon and a (four-coordinated)
aluminium. Again the oxide is affected by the Si 4+ cation and rsi equals a'l However,
the oxide is also affected by the A13+ cation for which rA1 = 1, ZA1 = 3 and hence
ZAlrA1/2 = g'3 Thus, by eq. (5):
X = 1 - (½(1 -- 1/7Si) + 3(1 - 1/7AI)},
where 7Si = 2.09 and TA1 = 1.65 (table 1), and X = 0.59. If the aluminium is six-coordinated instead of four-coordinated, rAl = l, ZAlrA1/2 = l, and X = 0.65.
Table 3 gives values of X for oxides in various bridging and non-bridging units encountered in oxide glasses. It should be noted that they were worked out on the as* Strictly the values are for glasses in which the alkali is lithium; differences in basicity in going
from one alkali metal to another are discussed later.
J.A. DullY, M.D. lngram / An interpretation of glass chemistry
382
Table 3
Microscopic optical basicities (theoretical) h for individual oxides in o x y a n i o n networks, etc. a).
Bridging or non-bridging oxide
BBBBB
BBBB-
O
O O O O O O O O-
B (3)
B (4)
Mg (6)
AI (4)
AI (6)
Si (4)
P (4)
H (1)
Microscopic optical basicity h b)
three-coordinate boron
four-coordinate boron
0.71
0.42
0.50
0.68
0.56
0.63
0.45
0.34
0.41
0.78
0.50
0.57
0.75
0.64
0.70
0.52
0.41
0.48
four-coordinate silicon
Si Si Si Si Si Si Si Si-
O
O
O
O
O
O
O
O
-
Si (4)
B (3)
B (4)
A1 (4)
AI (6)
P (4)
H (1)
0.74
0,48
0,45
0,52
0.59
0.65
0.36
0.44
four-coordinate phosphorus
P- O
P - O - P (4)
P - O - Si (4)
P - O - AI (4)
P - O - A1 (6)
P-O-H(1)
AI AI A1 AI -
0.63
0.25
0.36
0.48
0.54
0.32
four-coordinate a l umi ni um
six-coordinate a l umi ni um
O
0.85
0.91
O - A1 (4)
O - AI (6)
O - H (1)
0.70
0.77
0.55
0.77
0.83
0.61
a) Numbers in parentheses denote coordination numbers.
b) It is assumed for the purpose of calculation that all coordination positions are similar (see
text).
s u m p t i o n t h a t t h e b a s i c i t y m o d e r a t i n g p a r a m e t e r o f t h e n o n - p o l a r i z i n g c a t i o n s is
u n i t y . O w i n g t o t h e n e c e s s i t y o f h a v i n g t o c h o o s e c a l c i u m o x i d e as t h e e n v i r o n m e n t
p r o v i d i n g f r e e o x i d e s f o r t h e P b 2÷ p r o b e i o n ( s e e a b o v e ) , t h e l e s s e l e c t r o n e g a t i v e alk a l i a n d a l k a l i n e e a r t h i o n s will h a v e b a s i c i t y m o d e r a t i n g p a r a m e t e r s less t h a n u n i t y .
ZA. Duffy, M.D. Ingram / An interpretation of glass chemistry
383
If these metal ions are present instead of Ca 2+ or Li÷ ions, then many of the X-values
in table 3 will be increased. (This is usually not serious for Na ÷, but might be for,
say, Cs÷.) However, in this discussion we shall use X-values mainly for comparing
differences within one particular glass and not for comparing differences brought
about by substituting one alkali or alkaline earth ion for another.
Easteal and Udy [7] have also recently proposed the assignment of X-values to
different groups in borate glasses. While acknowledging that their procedure for calculating these group basicities differs from ours, we feel it appropriate to use the
symbol X to cover all aspects of local basicity in glass.
1.5. Internal neutralization effects and group basicities
The optical basicity concept embodies the principle that reactions between acids
and bases are those reactions which tend towards the equalization of optical basicity.
Thus it follows that the reaction between SiO2(A = 0.48) and Li20 (A = 1.00) is a
reaction between an acid and a base, because the product is a compound of intermediate basicity, e.g. Li4SiO 4 (A = 0.74).
Generally, it is not appropriate to regard a glass as a stoichiometric compound,
and indeed the chemical and physical properties are greatly influenced by the variations in microscopic basicity. The principle of equalization of optical basicity' still
operates in favour of those chemical and physical processes which tend to smooth
out variations in microscopic basicity. We are of the opinion that this principle is
the key to an understanding of many aspects of glass chemistry (and will be useful
even in glasses which are phase separated).
One immediate implication of the principle of equalization of microscopic basicity is that the crude microscopic basicities calculated in section 1.4 will have to be
modified whenever more than one type of oxide is present in an oxyanion group.
It is not clear to what extent the process of internal neutralization proceeds in such
a "mixed" oxyanion, although the difference in bond lengths [17] in, for example,
the SiO 4 unit of a metasilicate chain (fig. 5), where they are 1.68 and 1.57 A for
the bridging oxides and non-bridging oxides, respectively, suggest strongly that the
extent of internal neutralization is only partial. For the isoelectronic series SiO44- ,
PO43-, SO2-, C10~-, the microscopic optical basicity appears to vary smoothly with
bond length (fig. 6) and if one can use this relationship for estimating ;k from bond
0
0
0"
o-S~-o
0
"O.,*'S~-o 0
"o.S~.o 0t
I
(a)
I
(a)
I
(e)
Fig. 5. Groups representing the SiO 4 units of (a) orthosilicate ion, (b) metasilicate chain and (c)
silica.
384
J.A. DullY, M.D, Ingram / An interpretation of glass chemistry
Q,8
Q'7
X
Q.I~
-
Q,4 ~C10,0'3
I
1"5
I
1"6
J
1-7
~(A~
Fig. 6. Trend in microscopic optical basicity h with bond length r, for an isoelectronic series of
oxyanions. (Bond lengths from ref. [ 17] .)
length, then, by interpolation, ;~ for a non-bridging oxide in metasilicate will be
0.68 (compared with 0,74 for non-bridging oxides in the orthosilicate anion). (Unfortunately, there are no data for the isoelectronic series SiO 2, PO~, SO22+, C103+,
and it is not possible to obtain ~ for the bridging oxides in metasilicate.)
Thus it would s~eernthat the microscopic basicity of an oxide in a mixed oxyanion
unit lies betwee!l the value assigned on the basis of eq. (4) and the value if complete
neutralization occurred within the oxyanion unit. This latter value is referred to as
the group basicity, and is a very important parameter of a glass. As will be seen below, many properties can be explained by referring to the group basicity rather than
the microscopic basicity of the individual oxides.
Group basicities can be evaluated using eq. (4) (we shall continue to distinguish
fro'q the bulk basicity, A, and to use the symbol ;k for group basicity). Consider
for example the tetrahedral SiO 4 units in fig. 5. For the group in fig. 5c, all four
oxides are bridging, being attached to other silicon atoms. The central silicon, therefore, has a half share in four oxide ions and "owns" a total of two oxide ions; thus
rsi = ½ and zsirsi/2 = 1. Therefore ;k = [1 - 1 (1 - 1/2.09)] = 0.48. In the SiO4 unit
of fig. 5b two of the oxides are bridging and two are non-bridging, being terminal
oxides. Thus the central silicon owns three oxide ions, and with zsirsi/2 = 2/3, ~ =
0.65. Similar considerations and the use of eq. (4) allow values of the group basicity
to be calculated for other SiO 4 units (table 4). Table 4 also includes group basicity
values for various PO 4, SO4 and C104 units, and the values that are in bold type in
the table correspond to the unit that is present in the non-metal oxide, namely SiO 2,
P205 , SO 3 and C1207. In phosphate and sulphate glasses, only those tetrahedral
units having ~. values to the right of those that are in bold type are normally encountered; for example, a PO 4 unit where all the oxides are bridging would only occur in
a cationic PO~ unit.
J.A. Duffy M.D. Ingrain/An interpretation of glass chemistry
385
Table 4
Theoretical group basicities for tetrahedral oxy units a).
-O\
/OX
-O / \O-
X = Si
P
S
CI
0.48 b)
0.25
0.00
0.28
-O\
/O*
-O\
-O / \O-
/O*
X
- O / \O*
*O\ /O*
X
- O / \O*
*O\ /O*
X
*O/ \O*
0.58
0.40 b)
0.19
--0.02
0.65
0.50
0.33 b)
0.15
0.70
0.57
0.42
0.27 b)
0.74
0.63
0.50
0.36
X
a) Bridging oxide denoted O~ and terminal -O*.
b) Numbers in bold type correspond to the neutral oxide; numbers to the right of these are for
XO4 units occurring in oxyanions, while those to the left are for XO4 units occurring in (often imaginary) oxycations, e.g. PO~.
Table 5 gives group X-values for tetrahedral BO 4 units and trigonal BO 3 units. In
the tetrahedral BO 4 units the assignment of the group basicity values is less straightforward. Thus it is surprising to see the same group basicity value X = 0.57 being assigned to the
-
-o\
O\B-
O*
and
/OB
/
/\
-O
-O
O-
units, whereas chemical intuition suggests that the latter unit should be less basic
than the former. The reason for this apparent anomaly is that in the terminally
bonded oxygen all the "excess" negative charge is attached to one B 3+ cation and
the group is fairly well defined; in the tetrahedral boron atom, however, each bridging oxygen carries a formal charge o f - ¼.
Table 5
Theoretical group basicities for three-coordinated and four-coordinated borate units a).
three-coordinated boron
four-coordinated boron
-O
\B-O /
_O /
-O...
*O.
_O /
_O /
*O
"B-O*
*O/
A = 0.42
0.57
0.65
0.71
-%B/O-
-O
-O\
O* *O
O*
B/
\ B/
- O / \O* - O / "O*
*O.,
-O / \ O -
O*
B/
-O / " O -
A = 0.57
0.65
0.71
0.78
B-O*
a) Bridging oxide denoted -O" and terminal -O*.
B-O*
0.75
O*
B/
*O / "O*
386
J.A. Duffy, M.D. Ingram / An interpretation of glass chemistry
0"55
0"50
ApbX
0"45
0.40
I
0"1
I
1"0
I
1"5
Na20 : P20s ratio
Fig. 7. Plot of experimentally determined optical basicity, Apb(ii) , versus Na20 : P2Os ratio in
a ~odium phosphate glass system.
As stated above, the amount of internal neutralization which actually occurs
within the glass structure is difficult to quantify, but there are good grounds for
thinking that the probe ions (e.g. Pb 2÷) usually respond to group basicity rather
than to microscopic basicity. This is seen for example in the Na20-P205 glass system where the basicity, as registered by Pb 2+ ions, increases by 0.07 in the region
of the metaphosphate (I : 1) composition (fig. 7). Glasses more basic than NaPO 3
will contain a significant proportion of the terminal
i )
-O-
P - O*
I
O*
groups which will not be present in glasses less basic than NaPO 3. As can be seen from
table 4, the introduction of these groups should introduce a jump in group basicity
of 0.07 - in striking agreement with the measured rise in the experimental basicity.
2. Applications
2,1. Molar refractivitY o f O 2- ions
Reference must now be made to Weyl and Marboe's book [18] since a good deal
of what follows is based on our reinterpretation of some of their ideas in terms of
the optical basicity concept. Weyl and Marboe place great emphasis on the importance of the molar refractivity of the oxide ion RO2- (usually expressed in cm 3 mol-1).
They attribute changes in the refractivity of oxide, which are caused by coordination
of polarizing cations such as A13÷ and Si4÷, to a "tightening of its electron clouds".
A close correlation between molar refractivity and optical basicity is therefore to be
expected.
J.A. DullY, M.D. Ingram /An interpretation of glass chemistry
7"0
I
~
i
387
I
-2.5
6"0
s, s
I
E
ae
2-O
5"0
4-0
3"0
1"5
1
1"0
2-0
0"4
I
0"5
I
0"6
I
0"?
I
0"8
I
0"9
1-0
OPTICALBASIClTY
Fig. 8. Molar refractivity R and atomic polarizability c~ for various barium silicates (after Weyl
and Marboe [18], p. 63) plotted against theoretical optical basicity [using eq. (4)] for: 1 SiO2,
2 BaO : 2SIO2, 3 2BaO : 3SIO2, 4 BaO : SiO 2 and 5 2BaO : SiO 2.
Figure 8 shows molar refractivity data (from ref. [18]~ p.63) for a series of crystalline barium silicates plotted against values of A computed via eq. (4). The graph is
a good straight line, and clearly an increase in RO2- implies an increase in A. In
terms of the data included in fig. 8
A = ( R o 2 - - 1.5)/4.25.
(5)
Weyl and Marboe emphasize that the molar refractivity values of the 0 2 - ions
"are a summation over the responses of the electrons of all O 2- ions to an electrical
field", and point out that "the molar refractivity cannot help us in structural problems that make it necessary to differentiate between the polarizabilities of individual 0 2 - ions". Furthermore, it was the opinion of the authors that there was no
method for measuring the distribution of the values of RO2-- within a system. In
order to overcome this limitation they visualized ([ 18], p. 342) a qualitative scheme
such as that shown in fig. 9 which applies to a comparison of vitreous silica with
Number of
oxides per unit
volume of glass
~arbitrary)
(a)
,A4
3
I
L
5
6
(b)
3
4
5
6
Polarizability of oxide ~om3)
Fig. 9. Schematic diagram showing the polarizabilities of oxides in (a) vitreous silica, and (b)
alkali silicate glass (after [18], p. 342).
J.A. Duffy, M.D. Ingram / An interpretation of glass chemistry
389
ideas. Table 6 shows how k increases on going from the all-bridging situation in silica
to the all-non-bridging situation in the alkali orthosilicate, and it is apparent that the
difference in microscopic basicities ;k increases significantly from Li to Cs. Of course,
for an alkali silicate where the alkali oxide/SiO 2 ratio is less than 2 : 1, owing tO iriternal neutralization effects (see above) in the SiO 4 units containing both btidgfrlg
and non-bridging oxides, the value o f ~k will be less than the values indicated in
table 6.
It is necessary to be circumspect before using differences in 8k or k that arise as
a result of changing from one alkali metal to another. The reason for this is the possible unreliability o f the alkali metal 7-values, which may arise because o f the unrefined nature o f Pauling electronegativity in this part (group I) o f the Periodic Table
Since Pauling [21 ] put forward his electronegativity values, several other electronegativity scales have been proposed. Some of these appear at first sight to yield more
refined electronegativity values, and we now turn to one of the most popular, namely the Allred and Rochow [22] scale, as an alternative means o f calculating basicity
moderating parameters for the alkali metals. With the Allred and Rochow electronegativities the new basicity moderating parameter 3'(A.R.)* is given by
3,(A.R.) = 1.29 [x(n.R.) - 0.27] .
Values of 3,(A.R.) for the alkali metals, calculated from this equation, are given in
table 7, and can be compared with the original 7-values based on Pauling electronegativity. We consider presently which set of "r-values is better in reflecting the
polarizing action of the alkali metal ions in glass.
Weyl and Marboe give some emphasis to Dietzel's [23] use o f the field strength
(z/a 2) of a cation at distance a from the centre o f an adjacent oxide, and they prefer
this to the more conventional z/r 2 (r = cation radius). They state that z/a 2 can be
Table 7
Basicity moderating parameters 3", and field strengths z/a 2 of the alkali metal ions
Alkali metal
x(A.R.) a)
3' (A.R.) b)
3' (Pauling) b)
z/a 2 c)
Li
Na
K
Rb
Cs
0.97
1.01
0.91
0.89
0.86
0.90
0.95
0.83
0.80
0.76
1.00
0.87
0.73
0.73
0.60
0.23
0.17
0.13
0.12
0.10
a) Electronegativity values are those of Mired and Rochow.
b) Basicity moderating parameters 'r (A.R.) calculated from Allred and Rochow electronegativity values (see text); 3'-values obtained from Pauling electronegativities are given for comparison.
c) Values of z/a 2 are those pertaining to the alkali metal monoxides.
* Based on values of x(A.R.) for Ca and H, see ref. [14] for details of calculations.
J.A. Duffy, M.D. Ingram /An interpretation of glass chemistry
390
Li
1"0
Li
/
(a)
///
f
,"
(b)
Na
O
/
/
//
z/a 2
z/a'(Li)
/
/
/
/
oK
ORb
QOs
/
/
/
/
/
/
OK
ORb
OOs
/
p.
/
/
/
/
/
/
/
/
0
/
[
/
/
/
o
ONa
/
/
/,
• J
1"0
(A,n.)
i
1"0
0
~ ti
~[ Li(A'R')
Fig. 10. (a) Comparison of field strengths (z/a 2) with Allred and Rochow basicity moderating
parameters ['r(A.R.)] of the alkali metals: plot of (z/a2)/(z/a2)Li versus T(A.R.)/T(A.R.)L i. (The
dashed line is of unit slope.) (b) Comparison of field strengths (z/a 2) with original (Pauling)
basicity moderating parameters T of the alkali metals: plot of (z/a2)/(z/a2)L i versus 7/~'Li" (The
dashed line is of unit slope,)
used for estimating the interaction between the cations and oxide ions "and with
it the acidity or basicity of the oxide". This implies that the field strength of a cation is effectively the same as the basicity moderating parameter (suitably scaled).
However, because field strength involves the oxidation number z, it is only possible
to compare values of z/a 2 with T-values for unipositive ions.
When this comparison is made for the alkali metal ions (by normalizing all values
Li
1"0
p
p
p
I
s
b p
I
z/r 2
p
'/"C'i)
i I
/
i /
/
/
/
I A"
P
Fig. 11. Plot of
of unit slope.)
(z/r2)/(z/r2)Li
b
I
I
P
P
oNa
o,8,'b
i
1'0
?; L!
versus 3,/'YLi using original (Pauling) ")-values. (The dashed line is
J.A. Duffy, M.D. lngram / An interpretation of glass chemistry
391
to unity for lithium), it is found that the basicity moderating parameters obtained
from the Allred and Rochow electronegativities do not correlate at all well with the
field strengths (fig. 10a). As can be seen from fig. 10b, a much better (linear) correlation is obtained using the original "/-values (obtained from Pauling electronegativities). (lncidentaly, it is interesting to note in relation to Weyl and Marboe's preference for z/a 2 over z/r 2 that the 7-values correlate very badly with values ofz/r 2, see
fig. 11.)
An area where the effects of different alkali metal ions on optical basicity have
been investigated is in the alkali oxide - B203 glass system, and as we shall see below, data for this system obtained by Easteal and Udy [7] correlate well with the
original (Pauling) 7-values of the alkali metals. As discussed previously, owing to
the tendencies of probing indicator ions to show slight site selectivity, experimental
values of A do not tally exactly with the theoretical values. However, for a given indicator ion, it is possible that differences in optical basicity may become apparent,
at a particular alkali oxide content, on changing from one alkali metal to another.
At low alkali oxide compositions, differences in measured optical basicity are very
small and are thus subject to uncertainties in the experimental determination of
the 1S0 ~ 3P 1 frequency of the Pb 2+ indicator ion. However, at higher alkali oxide
contents, the optical basicity differences are larger and are therefore more reliable.
The greatest difference in optical basicity, for a particular alkali oxide content, will
be on going from the L i 2 0 - B 2 0 3 glass to the C s 2 0 - B 2 0 3 glass. Table 8 compares
the differences in experimentally determined optical basicity (for alkali oxide contents of 25%, 30% and 35%) with differences in theoretical A obtained from (i) the
original 7-values, and (ii) 7-values derived from Allred and Rochow electronegativities. The two sets of theoretical 3. differences are quite distinct from each other,
and comparison with experimental data indicates that the "Pauling" 7-values (table
1) for lithium and caesium express a numerical difference which accords with the
experimental basicity difference between lithium borate and caesium borate glasses.
In contrast, the T-values derived from Allred and Rochow electronegativities predict a difference between lithium and caesium which is much too small.
Table 8
Differences in optical basicity between Li20-B203 glass and Cs20-B203 glass.
Glass composition
25% alkali oxide
30% alkali oxide
35% alkali oxide
Optical basicity difference
Experimental
[7]
calculated
(Pauling)
calculated
(Allred and Rochow)
0.070
0.081
0.102
0.067
0.085
0.104
0.021
0.025
0.031
392
J.A. Dully, M.D. lngram / An interpretation o f glass chemistry
/
<
i
sO
0.10
sS
s
/
s"
,~
s S
0"05
J
J
s
/
/
/
/
ff
s
I
I
I
0
0.05
0.10
Dirl'el'en~ in A (th~retleal)
Fig. 12. Plot of experimental and theoretical differences in optical basicity for alkali oxide-B203
glasses (25%, 30% and 35% alkali oxide). Differences are between: Li-Cs ®, L i - R b A, L i - K ×,
Li-Na [] and Li-Li e. (The dashed line is of unit slope.)
To investigate how appropriate are the original 3,-values of the other alkali metals,
optical basicity differences between lithium and the alkali metals were calculated
and the results compared with experimental differences. As seen from fig. 12, the
original 7-values do express suitable numerical differences between all the alkali metals except for rubidium. Because potassium and rubidium have the same Pauling
electronegativity they therefore have the same basicity moderating parameter,
whereas the alkali borate glass data, and also the field strength data given in table 7,
would indicate a "r-value for rubidium which is slightly below that for potassium.
2. 3. Metal ions dissolved in glass
Many metal ions dissolve readily in molten silicates (borates, phosphates, etc.)
and are held in solution when the melt is quenched to a glass. Many aspects of
solute-solvent behaviour of these glasses can be studied if the metal ion is from
the transition series, since the response of such ions to environment can be monitored
by various spectroscopic techniques, the most obvious being absorption spectroscopy. It is usually quite a simple matter to establish (i) the oxidation state of the
metal ion that is favoured by the glass, and (ii) the stereochemical environment
that it provides for the metal ion. Both are affected profoundly by the optical basicity of the glass.
When a metal ion is dissolved in a glass, the oxide ions behave as donor atoms
and impart some of their electron charge cloud density to the metal ion. In some
glasses discrete complexes are produced. For example, it was shown [24] that in
KNO 3 - Ca(NO3) 2 glasses the Co 2+ ion exists as the complex ion [Co(NO 3)4] 2-.
However, nitrate glasses may be regarded as "non-network" glasses [25] and hence
J.A. Duffy, M.D. lngram / An interpretation of glass chemistry
393
contain discrete anions which are free to enter the coordination sphere of a metal
ion. In conventional network glasses, it is very difficult to define the coordination
sphere and possibly it is best to regard the metal ion as occupying a site in the glass
or as simply solvated by the glass.
If we have a metal ion capable of existing in two "medium" oxidation states, +2
and +3 say, then the +3 state will be favoured as the optical basicity of the glass is
increased, since a high A-value means that the oxides are able to donate more negative charge, thereby stabilizing the metal ion. This phenomenon is observed for
Fe2+/Fe 3+ and for Mn2+/Mn 3+ redox pairs in alkali borate glasses: as the alkali
oxide content is increased [thus increasing A (fig. 4)] the upper/lower oxidation
state ratio is also increased [26,27].
Some metals are more readily oxidized to what may be described as "high" (as
opposed to medium) oxidation states, e.g. +5, +6. The oxidation of chromium(Ill)
to chromium(VI) (by atmospheric oxygen) was studied as a function of alkali oxide
content in the alkali borate glass system [28]. In fact the chromium(IlI) - chromium(VI) equilibrium was used as a basis for one of the earliest optical indicators [1,
29,30] for the degree of glass basicity. In the conversion of chromium(Ill) to chromium(VI), it is necessary that the oxides have sufficiently high optical basicity that
at least some of them can detach themselves completely from the glass network.
This part of the mechanism can be envisaged as follows:
\
\
0
-~
0
->
-'0
" ~ Or(ill) t
0/
\
0
0
\o/i\o/1\ol
~0
0~
Cr(Vl)
o
-
+ 3e .
(a)
Since the redox half equation which balances the above reaction is
3 0 2 + 3e -+ ~3 0 2-
(b)
it might appear (from Le Ohatelier's principle) that the oxidation of chromium(Ill)
to chromium(VI) would be favoured by a decrease in the alkali content of the glass.
That this is not the case illustrates the overwhelming importance of glass basicity in
providing the correct environment for stabilizing a particular oxidation state.
In eq. (a), only part of the coordination spheres of chromium(Ill) and chromium
(VI) has been considered. As far as chromium(Ill) is concerned, it is probably best
to envisage, as mentioned above, a solvation of the metal ion by the glass, and this
would involve presumably both bridging ( X - O - X ) and non-bridging ( X - O ) oxides
as donor atoms. However, in considering chromium(VI) (or any other metal ion in
an "upper" oxidation state), the question arises as to whether the oxides, to which
the metal ion is attached, are separate from the glass network (as eq. (a) implies) or
whether one or more of the oxides are still attached to the glass network [thereby
acting as a bridge between X and Cr(VI): X - O - C r ( V I ) ] . Spectroscopic studies of
394
J.A. DullY, M.D. Ingram / An interpretation of glass chemistry
Na20 40
32
24
16
8
0
Fig. 13. Basicity c o n t o u r s (labelled a c c o r d i n g to A) in the glass s y s t e m L i 2 0 - N a 2 0 - l ~ 2 0 3 .
Critical basicities are d e n o t e d . . . .
for Co 2+ and - - for Cr 6+. (After refs. [ 1 6 , 3 1 , 3 2 ].)
chromium(VI) in alkali borate glasses indicate that there is an equilibrium between
these two coordinations [31,32] which depends upon the basicity of the glass. The
equilibrium has been used for constructing "isobasicity" lines in various ternary
systems such as Li20-Na20-B203, L i 2 0 - K 2 0 - B 2 0 3 and N a 2 0 - K 2 0 - B 2 0 3
[32]. These isobasicity lines follow closely the theoretical basicity contours which
can be constructed for these systems using eq. (4) (e.g. see fig. 13).
In ultra-high alkali borate glass (67Na20 : 33B203) and in sulphate glass
(40K2SO4 : 60ZnSO4) it appears that the chromium(VI) exists as the discrete
chromate (CrO42-) ion [33,34]. A similar situation seems to exist for vanadium(V)
which is present as the discrete vanadate(V) ion in sulphate glass but not in bisulphate glass [34]. Thus, whereas the medium oxidation state metal ions enter into
sites and are bonded into the network, it appears the upper oxidation state ions
have the choice of existing either as the discrete oxyanion, or of being partly attached to the network by one (or perhaps more) oxides that serve as a bridge.
Glasses with the highest optical basicity so far studied [2] are the ultra-high
alkali borate glasses where the Na20 content is in the range 66.5-71.5%. This composition range is very close to the composition corresponding to the stoichiometry
of the pyroborate ion B2O4-, and indeed the behaviour of the Co 2+ ion in this
glass [33] appears similar to that in nitrate [24] and acetate [35] glass, indicating
that probably the glass is of the non-network type. The glass with 68% Na20 is
found to have a A-value of 0.68. Not surprisingly, the upper oxidation states are
very much favoured, vanadium existing as vanadate, chromium as chromate (see
above) and manganese even existing as the blue manganate(V), MnO43-, species [33]
[and green manganate(VI), MnO2-, in the melt [36]].
Although optical basicity is the main factor in determining the correct environment for a particular oxidation state, it is nevertheless possible to carry out redox
reactions in the molten glass by controlling the atmosphere. In fact most glasses
seem to impart a redox behaviour to metal ions which is very similar to that of
J.A. Dully M.D. Ingrain/An interpretation of glass chemistry
395
water (which has a A-value of 0.40): oxidation states that are difficult, but possible,
to obtain in water are similarly so in glass, for example Ti 3+, V 3+.
Metal ions for which a change in stereochemistry is possible, also respond in an
interesting way to the optical basicity of the glass. Cobalt(lI), for example, in many
glasses exists in the pink octahedral form. As the basicity of the glass increases, however, there is a gradual change to the blue tetrahedral form [37,38]. In terms of
optical basicity, the explanation is straightforward in that as the microscopic k-values
of the oxides increase, so fewer are required to "neutralize" the Co 2+ ion and the
coordination number falls from six to four. As a rough guide, cobalt(II) is octahedral
when A < 0.5, tetrahedral when > 0.5. It is worth pointing out in this connection
that the colour of the glass is not always a good indication of whether cobalt(II) is
octahedrally or tetrahedrally coordinated [39].
Now although glass basicity is a major factor in determining the coordination
number of a dissolved metal ion, other variables such as temperature and pressure
can be important also. Indeed, Angell [40] discussed the use of transition metal
ions as probes for structural change in glasses and melts in relation to such thermodynamic concepts as free volume and configurational entropy.
The response of the vanadyl VO 2+ ion to glass basicity is also interesting since
this ion can register its sensitivity to the basicity of the environment through changes
in the degree of covalent interaction in the V - O bond of the vanadyl group. This
interaction is reflected in the frequency of the electronic transition of the single
d-electron to the antibonding 7r*-orbital associated with the V - O bond: as the
V - O bond strength increases, the n bonding orbitals decrease in energy while the
16000
14 000
12000
~
)
10OOO
t
B
t
i 160t~
~ 14000
12 000
10000
0"40
I
I
0-45
0.50
)-55
Optical llaslcity
Fig. 14. Correlation between theoretical optical basicity (a) of Li20-B203 glass system and (b)
of Na20-B203 with V-O frequency of VO2+ ion in these glasses. (The V-O frequency is that
of the transition 2B2 ~ 2Err).
396
J.A. Duffy, M.D. lngram / An interpretation of glass chemistry
corresponding antibonding 7r* orbitals increase in energy, thereby increasing the
energy separation between the latter orbitals and the orbital accommodating the delectron. Therefore, as the donor power (Lewis basicity) of the environment decreases, the greater is the extent of donation by the oxygen to the vanadium in the
V - O group. Lowering the basicity of the environment therefore brings about an
increase in the separation between the d-electron and the antibonding 7r*-orbitals
and hence an increase in the frequency of the absorption band. This correlation
was previously observed for the H2SO4-H20 system not only for VO 2+ but for
the analogous (4d 1) MoO 3+ species [41 ]. From the above reasoning it would be
anticipated that for the alkali oxide-boric oxide system, the vanadyl ion would
show an frequency decrease in this absorption band with increasing alkali oxide
content. This has been observed [42], and it is interesting to plot the frequency
shift against the optical basicity of the glass. For the Li20-B2G 3 system and especially the Na20-B203 system the relationship between A and the V - O frequency is almost linear (fig. 14) [16]. The VO 2+ ion appears in these two glass
systems to be sensing Lewis basicity in a manner similar to that of a dl°s2 probe
ion such as Pb 2+, and is in fact giving a "better" experimental verification of the
trend in theoretical optical basicity.
2.4. UV transparency
The UV transparency of a glass is usually impaired by even trace quantities of
metal ion impurities. Most (coordinated) metal ions have strongly absorbing bands
in the UV region, and it is the lower frequency edge of the first band which sets
the frequency limit to which the glass is transparent. Metal ions can be divided into
two types as far as the mechanism of UV absorption is concerned: (i) ions for which
the electronic transition involves orbitals located essentially on the ion itself, and
(ii) ions where the transition involves transference of an electron from oxide to the
metal ion. When the mechanism is of the first type, the absorption band is usually
influenced directly by the optical basicity of the glass, but for the second mechanism
the relationship between band position and optical basicity is less straightforward.
Metal ions for which the first mechanism operates include some rare earth ions
[43], but chiefly the p-block metal ions having oxidation numbers two units less
than their group number, e.g. Pb 2+ and Bi 3+. By virtue of the relationship between
optical basicity and the frequency of the first absorption band, it is possible to estimate the extent of the red shift the absorption band will suffer as the optical basicity is increased. These shifts are given in table 9 for T1+, Pb 2+ and Bi3+ [obtained
using eqs. (1)-(3)], and for In +, Sn 2+ and Sb 3+ (obtained using the equations applicable to these ions and derived from data in ref. [10]).
The second mechanism is much more common than the first. The energy of the
electron transfer process (and hence the position of the .,bsorption band) is determined by the electronegativRy difference between the metal ion and the glass. The
electronegativities are "optical electronegativities" ×, and the absorption maximum
J.A. Dully, M.D. Ingram /An interpretation of glass chemistry
397
Table 9
Frequency decrease of 1So --, 3P 1 absorption band for an increase of 0.1 unit of optical basicity
for p-block metal ions in glass.
metal ion
frequency decrease (cm-1)
Iin÷
T1÷
Sn2+
Pb 2+
Sb 3+
Bi3+
1100
1800
2300
3100
2500
2900
v(in cm -1) * is given by [8]
p = 30 000 (X (glass) - × (metal ion)).
(6)
× (metal ion) and × (glass) change on going from one metal ion to another or from
one glass to another. Studies in borate, silicate and phosphate glasses have indicated,
perhaps rather surprisingly, that aJtering the optical basicity of a glass does not necessarily bring about any corresponding change in the optical electronegativity of
the glass; it appears rather that it is necessary for the basicity to correspond to some
major change in glass structure for there to be any change in ×(glass) [44]. For example, the absorption band of Cu 2÷ in the N a 2 0 - B 2 0 3 glass system undergoes little
shift in the range 0 - 3 6 % Na20 , whereas in the N a 2 0 - P 2 0 5 glass system, ×(glass)
changes by nearly 0.1 unit around the 1 : 1 composition (where significant changes
in structure are known to occur) and correspondingly the band shifts by approximately 3000 cm-1 [44].
The UV transparency of glasses free o f trace metal ions improves as the alkali
oxide content, and hence the basicity, is decreased. The tightening of the outer electrons of the oxides, which is manifested by the lower X-values, results in increased
binding energy and hence the absorption o f light energy is shifted to higher frequencies.
Among the best glasses for UV transparency are the alkali and alkaline earth
phosphate glasses. These glasses have low optical basicities [-~0.46 - cf. the less
transparent N a 2 0 - S i O 2 (3 : 7) glass with A = 0.60] and when pure they have very
little absorption even at the UV frequency limit of commercial spectrophotometers
[just over 50 000 cm -1 (200 nm)]. However, since, as discussed above, the chief
cause of impairment of UV transparency in the glass is the presence of metal ion
impurities, it is important that the optical electronegativity difference [×(glass) x(metal ion)] should be kept as great as possible so that the electron transfer bands
absorb at the highest possible frequelacy [eq. (6)]. Thus the most important parameter, as far as common glasses are c oncerned, is the value of ×(glass) which should
• For certain ions, x must be corrected fo~: spin pairing and other effects.
398
J.A. Duffy, M.D. Ingram / An interpretation of glass chemistry
be as high as possible. Phosphate glasses, with a value of ~- 3.3 for ×(glass) are therefore superior in this respect compared with silicate and borate glasses which have
values of 3.15-3.25.
2. 5. Changes in the network coordination number
The structures of silicate glasses are relatively simple in that silicon is always
four-coordinate, and therefore the only groups which have to be considered are the
various SiO 4 tetrahedra depicted in fig. 5. In these groups the number of bridging
(and also terminal) oxides can vary from zero to four, and except in the case where
all the oxides are either bridging or non-bridging there will be an imbalance in microscopic basicity hk. This imbalance is only partially diminished by the internal neutralization, previously referred to, within the SiO 4 tetrahedron, but in other glasses
variations in microscopic basicity can be reduced by changes in the coordination
number of the network-forming cation.
This is the situation pertaining in glasses containing B 3+, A13+ and Ge 4+, but we
shall consider as examples only the alkali borate and alkali aluminosilicate glasses.
We take as our working hypothesis that changes in coordination number occur, if
by doing so the difference in microscopic basicities/Sk between 0 2 - ions attached
to a common network-forming cation is thereby reduced (see above).
Because of the ability of boron to change between three- and fourfold coordination, there are two possible reactions that can occur when alkali oxide is added to
B203 (fig. 15). Retention of threefold coordination by the boron requires the formation of non-bridging oxides within the BO 3 units. However, if some borons
change their coordination number to four, the added 0 2 - ions are incorporated as
bridging oxides, and the creation of non-bridging oxides is avoided. In changing
from a BO 3 to a BO 4 unit (fig. 15b), a bridging oxide undergoes an increase in
01
0S
I
I
~,0 ..."B "~0 4. B,,,.O t
+ 0 2-
/
0t
I
.o,B.o
0"
I
÷
(a)
o.B.o.
%
0"
I
0"
I
O,B O2B o(b)
Fig. 15. Addition of alkali oxide to B20 a to produce (a) non-bridging oxides on BOa units, or
(b) BO4 units (but no non-bridging oxides).
J.A. DullY, M.D. Ingram /An interpretation of glass chemistry
399
ACIDIC
BASIC
0.8
J
.~ 0"?
o 0"6
.u
.S)-_O-_~
0-5
.|
Si-O-Si i
• 0.4
I
-- Si-O..AI
I
0"3
/
MzO/AI20~ Patio >1:1
I
I
,,
1:1 <1:1
Fig. 16. Spread in microscopic optical basicity, 6X (~ arrow length), for different Na20/AI203
ratios, in sodium aluminosilicate glasses. (Silicon and aluminium are both assumed to be fourcoordinated.)
from 0.42 to 0.57 and ~5~.is 0.15. If terminal oxides are produced (fig. 15a), X increases from 0.42 to 0.71 and 6X is 0.29. In accordance with our hypothesis., it is
observed experimentally that addition of alkali to borate glasses does favour the
production of four-coordinate boron rather than the formation of non-bridging
oxides [45,46]. Furthermore, it would be expected that the reaction would not be
as simple as that depicted in fig. 15b, but would yield not BO 4 units adjacent to
each other but BO 4 units that were linked entirely to BO 3 units. As seen from table
2, an oxide bridging a four-coordinate boron to a three-coordinate boron has ~ =
0.50 and ~ is then only 0.08.
It is a generally accepted principle [47] (see also ref. [ 18, p.510[) that the proportion of non-bridging oxide in sodium aluminosilicate glasses decreases with increasing A1203 : Na20 ratio, and that when this ratio is unity all the oxygens are
in bridging positions. The [A104]- groups are thus the characteristic structural
entities in these glasses. However, when the AI203 : Na20 ratio is greater than unity
then the coordination number of oxygen must rise, and there has been some discussion in the literature as to existence of A106 octahedra [48,49] and A104 triclusters [50,51].
The spread in the microscopic basicity ~ in sodium aluminosilicate glasses for
different A1203 : Na20 ratios and assuming retention of fourfold coordination is
shown schematically in fig. 16. (The values are taken from tables 3 and 10 which
strictly refer to Li-containing glasses, but in view of the internal neutralization effects it is hardly worth calculating new values here for each alkali metal ion.) The
main feature in this diagram is that the range in microscopic basicities/SX goes
through a minimum at the A1203 : Na20 ratio of 1 : 1. This is in accord with the
notion that this composition marks some kind of "neutralization point" and has
an important bearing on the ease of workability and resistance towards devitrification of glasses around this composition.
J.A. Duffv, M.D. Ingram / An interpretation of glass chemistry
400
Table 10
~,-valuesof three-coordinate oxides in aluminosilicate networks.
four-coordinate AI
six-coordinate
AI
Si
I
O
Si/ \A1
0.33
0.38
Si
0.44
0.54
0.55
0.70
I
AI/O\AI
A1
I
AI/OA1
Values of k for various types of three-coordinate oxygen in aluminosilicate glasses are given in table 10. This list is not exhaustive, but it shows that by making
suitable permutations among the possible arrangements of Si and A1 it is possible
to obtain 0 2- ions of almost any desired basicity. This is perhaps the clue as to
why the chemistry of aluminosilicates is so complex, and may ultimately explain
some of the puzzling changes in Al-coordination number which have been proposed [52]. In LiA1SiO4 glass it is believed [18, p. 514] that A13÷ ions are present
in sixfold coordination.
2. 6. Chemical durability and acid-base equilibria at the glass/water interface
Concerning the technology of glass this is obviously a topic of paramount importance. Stumes into the chemistry and physics of glass would lose much of their
point if glasses were not available to stand service in window panes, glass containers
and in precision optical instruments. The literature on this subject is extensive and
frequently in apparent conflict (the situation for metallic corrosion is similar), and
it is fortunate that Weyl and Marboe [18, pp. 1010 et seq.] have provided a comprehensive and lucid review of the field. It appears that glass durability depends on
many factors including glass composition, thermal history, pH of the corroding
medium and the presence of various catalysts and inhibitors. For silicate glasses,
the production of a surface layer of hydrated silicic acid has also been shown to be
very much involved in the mechanism of glass dissolution.
A full treatment of surface durability is obviously outside the scope of this paper.
However, two factors which can be singled out as contributing to a decrease in durability are: (i) increase in the number of non-bridging oxides in the glass, and (ii) increase in pH of the dissolving medium. The possible link between these two factors
may well be the Lewis basicity of the oxides in (i) the glass, and (ii) the aqueous
medium, and we therefore examine this problem in quantitative terms.
J.A. DullY, M.D. Ingram /An interpretation of glass chemistry
401
To tackle the problem of proton exchange between glasses and aqueous solutions,
we found it best to start with the well-documented behaviour of oxyanions in solution. In the context of the B r o n s t e d - L o w r y theory, all oxyanions may be regarded
as the conjugate bases of their parent acids, regardless of whether the acids are weak
or strong. Thus, for example, the NO~ ion is the conjugate base of HNO 3, and in
principle equilibria can always be written of the type
NO 3 + H3 O+ . . . . HNO 3 + H20 ;
For this equilibrium,
K' = (a(ltNO3) a(H20))/(aCNO~ ) a(H30+ ))
and log K' is obviously the pK of the conjugate acid (HNO3) (pK data for inorganic
acids are readily available in a standard reference text [53]). If proton affinity is
14
1.
2
12
HzO/OH-
2..Po:'/~o:-
E)
3. HP2O~"/ PzO~"
4. HzPO="
/HPO~z"
s. H,P~O~-/.P,O,"
10
6. H,P,O.~/ H2PtOv=-
3
8
7. H=PO,/ HzP04"
8. H4P2OT/H~P=O7"
9. HNO=/ NO;
10.H30+ / HzO
5
D.
6
4
21
0
BE)
0"3
0"5
0"6
0"7
~E
"2
i
A (CONJUGATE BASE)
Fig. 17. Plot of pK (oxyacid) versus the optical basicity of the conjugate base. (N.B. A for NO3
is 0.39 and not 0.50 as stated inref [14].) It should be noted that the pK-values of the H30+/H20
and H20/OH- acid-base couples are set to -1.74 and +15.74, respectively, and not to zero and
+14, since in aqueous solution the concentration of water is 55 M.
402
J.A. Duffy, M.D. Ingram / An interpretation of glass chemistry
determined by optical basicity, we shall expect to find a correlation between optical
basicity of the anion and the pK of the conjugate acid.
Of the principal glass-forming cations (Si4+, B 3÷ and p5+), the oxyacids of silicon
and boron provide very meagre data as far as their acid strengths in water are concerned (the equilibrium constant is complicated for boric acid because of the change in
coordination number: B(OH)3 + H20 ~ H+ + [B(OH)4] - and for silicic acid because
of condensation reactions). However phosphorus(V) forms, in addition to orthophosphoric acid, H3PO4, several condensed oxyacids containing both bridging and nonbridging oxides (this last feature makes them particularly attractive for the present
purposes) and among these is pyrophosphoric acid H4P207 for which all four dissociation constants have been obtained. In fig. 17, pK (conjugate acid) values for the
arious phosphate species are plotted against the optical basicity A and for the "bases" which are always present in water, namely H20 and O H - . The salient feature
of this graph is the linearity between pK and A. If the "best" straight line is taken
to be that which passes through the data points relating to the self-ionization of
water, then it follows that
pK + 1.74 = 58.3 [A - 0.401 .
(7)
In simple terms, an anion will attract protons, and hence the conjugate acid will be
weak, if its optical basicity is greater than that of water (A = 0.40). Incidentally, it
is found that the above relationship holds quite well for a much wider range of oxyacids which involve elements other than phosphorus [54], and the equation is therefore important in that it establishes for the first time a quantitative link between the
Bronsted theory of acidity for protonic systems and a theory of basicity which has
been developed entirely within the context of oxyanion melts and glasses. It cannot
be emphasized too strongly that the numerical values of pK and A used to establish
this correlation are entirely independent o f each other.
As far as the surface chemistry of glass is concerned two points may be noted.
First, the pK value of an acid depends only on A and is independent of the charge
carried by the conjugate base. This is a useful simplification. Second, the X-values of
the oxyanions are group basicities rather than microscopic basicities. The surface of
an oxyanion glass, wetted by an aqueous medium, can be regarded as a particular
"condition" of a hydrated anion. For the purposes of considering the protonation
at a glass/water interface, we can apply the same principles as for the oxyanion/water
interface. From this, it follows that the extent of protonation of a glass surface can
be estimated by calculating X-values corresponding to the "groups" which are present
in the surface layers, and obtaining pK values from eq. (7) or fig. 17 for these groups.
In table 11 are the group basicities X for several structural units likely to be present in glass surfaces together with the predicted values of pK. For a SiO 4 tetrahedron containing one non-bridging oxide, it is apparent that this oxide ion should
have a 750% chance of existing in a protonated condition at all pH values below
about 9. Above this pH-value non-bridging oxides will become progressively deprotonated, and it is significant to note that experimentally the "alkaline attack" on glass
manifests itself in this pH region.
£A. Duffy, M.D. Ingram / An interpretation of glass chemistry
403
Table 11
Theoretical group basicities h, and pK values lbr units in aluminosilicate glasses.
Group a)
h
pK b)
(Si)-O
/O*
"Si
(Si)-O / \ 0 (Si)
0.58 c)
8.8
(Si)-O\
0.65 c)
12.8
0.59 d)
9.3
0.52
5.2
/0"
Si
(Si)-O /
\0"
(Si)-O\AI'/O-(Si) (aluminosilicate
(Si)-O /
\O-(Si)
group(A))
"Al (SIO4)4" (aluminosilicate group (B))
a) Bridging oxides denoted --O" and terminal O*.
b) pK values obtained from h using eq. (7).
c) Values also given in table 4.
d) Same as Si-O-A1 (4) values in table 3.
Dissolution of a silicate glass (in alkaline solutions) must involve some breaking up
of the network structure, and this process must be preceded by hydration of the glass
surface. It is reasonable to assume that the areas that are most heavily hydrated (i.e.
those having the greatest affinity for water molecules) will be most prone to break
away from the network structure and go into solution. Such "hot spots" will exist
in regions where the oxides have significantly higher microscopic optical basicity
than average since a high optical basicity goes hand in hand with a high electron
density on the oxide, and it is this concentration of negative charge which serves
to attract the positive ends of the water molecules. Removal of a proton from a
non-bridging oxide (which occurs for pH >9, see above) leads to the formation of
one of these hot spots, and the situation can be envisaged as shown in fig. 18.
®
0 ~"0 "0~
(a)
high
(b)
Fig. 18. Pictorial representation of the hydration of an oxyanion unit at the surface of a glass
(drawn flat for clarity) when (a) 6 k is zero leading to uniform hydration, and (b) ~ ?, is large
leading to a hydration "hot spot". (Charge density represented schematically by hatching.)
404
J.A. Duffv, M.D. Ingram / An interpretation of glass chemistry
It is interesting to relate this theory of glass dissolution to the following situation
cited by Weyl and Marboe [ 18, p. 64] to illustrate the uselessness of bulk RO2values and the necessity of being able to "differentiate between the polarizabilities
of individual 0 2 - ions" (which the assignment of microscopic optical basicities X
enables us to do). Weyl and Marboe state that it is possible to lower the molar refractivity of a sodium silicate glass by adding either silica or alumina, but that "the
effects of the two additions upon the chemical resistance of the resulting glasses are
very different and do not parallel the molar refractivity". In considering this problem,
and in accounting for the greater chemical resistance (i.e. lower chemical activity) of
the aluminosilicate glass, particularly under alkaline solutions, compare an aluminosilicate glass of composition Na20-A1203-SiO 2 (1 : 1 : 6) with the sodium silicate
glass Na20-SiO 2 (1 : 7). Both glasses have the same Na20 content and their theoretical optical basicities are very close (0.54 and 0.52, respectively). With a Na20 :
A1203 ratio of ! : 1, assuming fourfold coordination for the aluminium (see above)
the Na20-A1203-SiO 2 (1 : 1 : 6) glass will contain no non-bridging oxides. Disregarding any AI-O-A1 groups (if they are present at all, their concentration would
be verysmall), oxides of highest basicity will be those in the A I - O - S i group which
have X = 0.59 (table 2). This compares with the oxides of highest basicity in the
Na20-SiO 2 (1 : 7) glass, which will be the non-bridging oxides having X = 0.74. The
majority of oxides in both glasses will be oxides bridging two silicons, for which
X = 0.48, and it therefore follows that for the aluminosilicate glass ~X is only 0.11
compared with 0.25 for the sodium silicate glass. Therefore, in the latter glass there
is a greater electrical imbalance between the oxides of the oxyanion units with the
production of hydration "hot-spots" and ensuing decomposition of the glass.
2. Z Theglasselectrode
The properties of H÷-selective glass electrodes have been discussed in monographs
[55,56] and in recent publications [57,58]. It has been established that protons
from solution invade the glass structure (followed by water molecules) leading to the
formation of a leached layer of hydrated silica gel which is effectively dealkalized.
The pH response is governed in part by the mobility of H+ ions in the bulk of the
glass membrane, but more significantly by equilibria at the gel/solution interphase.
Thus Baucke [57] has drawn attention to the importance of acid-base equilibria
of the type
SiOHgel + H2Osolution ~
+
SiOgel + n 3Osolution
From the data in table 11, it can be seen that a SiO 4 group containing one nonbridging oxide is assigned a pK value of 8.8, which means that at a pH of 8.8, the
tendency of this SiO4 group for existing in the protonated form is the same as in
unprotonated form. Therefore, at pH values greater than 8.8, the tendency for the
group to exist in the unprotonated form will progressively increase. Referring to the
actual performance of glass electrodes, it is indeed noteworthy that Coming 015
J.A. Duff v, M.D. Ingrain/An interpretation o f glass chemistr),
405
glass (Na20 : CaO : SiO 2 = 26 : 6 : 72) performs perfectly satisfactorily when pH <9,
but in solutions of higher pH shows the well-known "alkaline error".
From about 1940 [59,60], electrode glasses containing Li20 as the principal
alkali became available for H+-ion measurement. These glasses have certain advantages over the Coming O15 composition used previously, in particular their response
to Na + ions is much reduced. This is in part due to the existence of a mixed-alkali
effect which reduces the Na + mobility in the bulk of the glass membrane. However,
it is interesting to note that the "best" pH glasses are multicomponent mixtures
[59,60] including a few mol% of an "oxygen modifier" such as La203 "to provide
a high ratio of oxygen to silicon". Such glasses contain only about 63% tool% SiO 2,
and therefore some of the SiO 4 structural units must contain more than one nonbridging oxide. From table 11, it is seen that such groups have a pK value ~12, and
should therefore show a pH response over an increased range. Recent studies of pH
glass electrodes by Baucke [61,62] using ion-sputtering techniques [63,64] indicate
a pK value o f - 1 0 . 3 , and it is interesting to note that this is midway between our
calculated values in table 11.
The presence of m1203 in glass markedly enhances the alkaline error and leads to
a diminished range of pH response. Indeed, A1203 has been added deliberately to
glass compositions to render the glasses insensitive to pH so that instead they can
be used as cation-selective glass electrodes. The question naturally arises as to whether
or not this effect can be deduced from changes in the optical basicity of structural
entities in the glass surface. As stated above, the characteristic "structure" of aluminosilicate glasses envisages A104 tetrahedra fitted into the network of SiO 4 tetrahedra. However the assignment of a "group basicity" to the A104 units is not straightforward.
in table 11, two different versions of the AIO4 group, as it exists in aluminosilicate
glasses, are given. In the first (group A), the group comprises only those four oxides
directly bound to the central A13+. The group basicity ?~is the same as the microscopic basicity for each oxide in a S i - O - A I (4) unit (0.59). Further reference to table
11 shows that this value of ~ is actually greater than the basicity of the
/
O
-
O -
Si -
L
O*
0,,
group (0.58) and it is difficult to see why pH response should be impaired. However,
this version, group A (table 11), makes no allowance for internal neutralization within the four SiO4 tetrahedra linked by the A13+ ion. In other words, the A104 unit
(group A in table 11) should be considered not on its own but as part of a larger
group which incorporates the whole of the four SiO 4 units surrounding the A13+
406
J,A. DullY, M.D. Ingram / An interpretation of glass chemistry
(group B in table 11). The group basicity is now 0.52 and pK -- 5.2. According to
this figure, glass electrodes containing significant amounts of A1203 should fail to
respond to H ÷ ions at pH > 5. In fact, departure from pH response begins when
pH > 3, but in view of the difficulty of identifying all the entities present in ahiminosilicate glass (see above) this disagreement is not too discouraging.
3. Conclusions
3.1. Theories o f acid-base behaviour
Theories of acid-base behaviour in glasses fall conveniently into three main classes. These are as follows: (i) Lux-Flood approach [65,66] ; (ii) self-ionization theories [67]; (iii) optical (Lewis) basicity concept [5].
According to the Lux-Flood theory, the characteristic acid-base reaction involves
exchange of 02_ ions, and we can write: base ~- 0 2 - + conjugate acid.
It is true to say that this theory has won general acceptance in the treatment of
acid-base equilibria in molten salts and slags, as well as in glasses. Indeed, following
on from the classic paper of Flood and Forland [66] it has become almost axiomatic
that the correct expression for the basicity of an oxyanion system is in terms of an
oxide ion activity, a02_, and various attempts have been made to set up pO 2- scales
analogous to the pH scale applicable in aqueous solutions. However, this procedure
is not free from ambiguity. Thus from the standpoint of rigorous thermodynamics
various authors [68,69] have cited Guggenheim [70] as to the indefinability of
single ion activities. Attempts to use the concept of oxide ion activity have led to
at least two controversies in the glass literature in recent years [68,69, 71-73[.
In view of the popularity of the a o 2 - concept, it is easy to forget that the 0 2 ion has no independent existence. Thus, the free 0 2 - ion in the gas phase, if it
existed, would have zero ionization energy and infinite polarizability [74]. In
chemical environments, O 2- is stabilized only by interaction with acidic cations
such as Ca 2+ or H +. Its polarizability is thereby reduced and becomes comparable
with those of other anions. If this is borne in mind, then there is no difficulty in
accepting the finding of Zambonin and Jordan [75] that addition of Na20 to molten
alkali nitrates leads to the formation of 0 2 - and O~- ions and very little 0 2 - . It is
now obvious that the controversy which surrounded these important results for
several years was largely due to the general acceptance of the notion of oxide ion
activity.
It remains to be asked whether or not the Lux-Flood theory can be modified
so that its application to glass chemistry becomes more fruitful. Flood and Forland
[66] define an acid-base reaction as one in which "an oxide ion changes from one
state of polarization to another". It is our opinion that this is a correct and entirely
definitive statement of the problem, but that their assumption that this "state of
polarization" can be expressed in terms of oxide ion activity is not valid, and cer-
J.A. Duff.v, M.D. lngram/ An interpretation o f glass chemistry
407
tainly does not apply to oxides in glass networks *. Following the arguments set
forth earlier in this paper, it is possible to unify both the L u x - F l o o d and Lewis
theories by regarding the state o f polarization of oxide ions in terms o f optical
basicity. Thus we restate the L u x - F l o o d concept as it should be applied in glass
chemistry as follows: " A n acid-base reaction in an oxyanion glass is one in which
oxide ions change from one state of (optical) basicity to another". It will be recalled, as previously discussed, that equilibria involving glass surfaces and anions
in solution can also be treated in terms of optical basicity changes; therefore this
statement holds for both protonic and non-protonic systems.
Self-dissociation theories are usually thought of as elaborations o f the L u x Flood concept in that the basicity of a particular glass is measured in terms o f
some specified 0 2 - exchange reaction. Thus Paul [67] has argued that basicity in
borate glasses depends on equilibria of the type
-
-O
B\
-'~ - B + * O - B - . .
I
However, it is apparent that free oxide ions do not have to be present in the glass
or melt for a c i d - b a s e reactions of this type to occur. This approach is attractive in
that it gives a structural interpretation o f acid- base behaviour, but it seems more
satisfactory to base the quantitative treatment not on oxide ion activities but on
group or microscopic basicities.
In view o f the above discussion the important question arises as to why so little
attention has been given to applying the Lewis concept of a c i d - b a s e behaviour in
glasses and related systems. In the past there seem to have been two major difficulties. First, the concept o f Lewis a c i d - b a s e interaction has been envisaged in
terms of reactions involving completion of the Lewis octet [18, p. 55] and generally Lewis acids and bases are thought of in terms o f species such as BF 3 and NH 3.
Second, with the Lewis approach there was the lack of a quantitative scale of acid
or base strength [76].
The concept of optical basicity removes these difficulties by recognizing that
the interaction between a probe metal ion and the oxides in a glass system is a
Lewis a c i d - b a s e interaction and that the degree o f electron donation can be
measured by UV spectroscopy. With the increased understanding of ionic interactions which have been obtained recently, it is now possible to use probe ions to
detect and measure the donor abilities of 0 2 . ions in any glass network. This is
* These comments should not be taken to imply the general uselessness of oxide ion activity
data. In certain molten salts containing discrete oxyanions, e.g. carbonates and nitrates,
oxide can be added to form solutions containing well-defined solute species such as O 2-,
OH- and 0 7. In these cases, a full description of acid-base behaviour requires a combination of the optical basicity concept with the more traditional equilibrium constant approach
as discussed elsewhere [141.
J.A. DullY, M.D. h~gram / An interpretation o f glass chemistry
408
the logical f o u n d a t i o n of the optical basicity concept, which appears to us to be
a " n a t u r a l " approach to glass chemistry, and one moreover to which other workers
are making i m p o r t a n t c o n t r i b u t i o n s [ 7 , 7 7 , 7 8 ] . It is, however, the correlation between optical basicity and Pauling electronegativity which is especially useful. The
correlation has made possible the calculation o f microscopic and group basicities
which are used extensively in this paper. As well as providing a scale of basicity for
systematizing the behaviour of" various indicator ions in glass, optical basicity provides
a convenient framework for the quantitative application o f electronegativity to a wide
range of chemical problems.
Acknowledgement
We t h a n k Dr. F.G.K. Baucke, Dr. Easteal and Dr. Udy for allowing us to use
hitherto unpublished results.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[ 18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
W.A. Weyl and E. Thtimen, Sprechsaal 66 (1933) 197.
J.A. Duffy and M.D. Ingrain, J. lnorg. Nucl. Chem. 36 (1974) 43.
J.A. Duffy and R.J. Grant, J. Phys. Chem. 79 (1975) 2780.
J.A. Duffy and M.D. Ingram, J. Chem. Phys. 52 (1970) 3752.
J.A. Duffy and M.D. Ingram, J. Am. Chem. Soc. 93 (1971) 6448.
A. Paul, Phys. Chem, Glasses, 11 (1970)46; 13 (1972) 144.
A.J. Easteal and D.J. Udy, Phys. Chem. Glasses, 14 (1973) 107, and personal communication.
C.K. J~rgensen, Absorption spectra and chemical bonding in complexes (Pergamon Press,
Oxford, 1962); Oxidation numbers and oxidation states (Springer-Verlag, Berlin, 1969).
J.A. Duffy and M.D. Ingram, J. Chem. Phys. 54 (1971) 443.
J.A. Duffy and M.D. lngram, Inorg. Claim. Acta 7 (1973) 594.
J.A. Duffy and M.D. lngram, Phys. Chem. Glasses 15 (1974) 34.
J.A. Duffy and M.D. Ingrain, J. Inorg. Nucl. Chem. 36 (1974) 43.
J.A. Duffy and M.D. Ingram, J. Chem. Soc. Chem. Commun. (1973) 635.
J.A. Duffy and M.D. Ingram, J. Inorg. Nucl. Chem. 37 (1975) 1203.
J.A. Ouffy and M.D. Ingrain, J. Inorg. Nucl. Chem. 36 (1974) 39.
J.A. Dully and M.D. Ingram, Phys. Chem. Glasses 16 (1975) 119.
A.F. Wells, Structural inorganic chemistry, 3rd ed. p. 57. (Oxford University Press, 1962)
p. 57.
W.A. Weyl and E.G. Marboe, The constitution of glasses (Interscience, New York, 1962).
J.R. Tessman, A.H. Kahn and W. Shockley, Phys. Rev. 92 (1953) 890.
B.E. Dempsey, Molecular sieves (Society of Chemical Industry, 1968) p. 293.
L. Pauling, The nature of the chemical bond, 3rd ed. (Cornell University Press, 1960) p.93.
A.L. Allred and E.G. Rochow, J. Inorg. Nucl. Chem. 5 (1958) 264.
A. Dietzel, Glastech. Ber. 22 (1948-1949) 41, 81,219.
J.A. Duffy and M.D. Ingrain, J. Am. Ceram. Soc. 51 (1968) 544.
M,D. Ingrain and J.A. Duffy, J. Am. Ceram. Soc. 53 (1970) 317.
J.A. Duff.v, M.D. Ingrain/An interpretation of glass chemistry
409
[26] A. Paul and R.W. Douglas, Phys. Chem. Glasses 6 (1965) 207.
[27] A. Paul and R.W. Douglas, Phys. Chem. Glasses 7 (1966) 1.
[28] P. Nath and R.W. Douglas, Phys. Chem. Glasses 6 (1965) 197.
[29] H. Lux and E. Rogler, Z. Anorg. Allg. Chem. 250 (1942) 159.
[30] W. Stegmaier and A. Dietzel, Glastech. Ber. 18 (1940) 247.
[31] A. Paul and R.W. Douglas, Phys. Chem. Glasses 8 (1967) 151.
[32] A. Paul and R.W. Douglas, Phys. Chem. Glasses 9 (1968) 27.
[33] J.A. Duffy, Phys. Chem. Glasses 16 (1975) 22.
[34] J.A. Duffy and W.J.D. MacDonald, Phys. Chem. Glasses 11 (1970) 1.
[35] M.D. Ingram, G.G. Lewis and J.A. Duffy, J. Phys. Chem. 76 (1972) 1035.
[36] W.D. Bancroft and R.L. Nugent, J. Phys. Chem. 33 (1929) 481.
[37] M.A. Aglan and H. Moore, J. Soc. Technol. 39 (1955) 351T.
[38] A. Paul and R.W. Douglas, Phys. Chem. Glasses 9 (1968) 21.
[39] M.D. lngram and J.A. Dully, J. Chem. Soc. A (1968) 2575.
[40] C.A. Angell, J. Am. Ceram. Soc. 51 (1968) 125.
[41] J. Arris and J.A. Duffy, J. Chem. Soc. (1964) 5850.
[42] H. Toyuki and S. Akagi, Phys. Chem. Glasses 13 (1972) 15.
[43] R. Reisfeld, Structure and bonding (Springer-Verlag, Berlin, 1973) p.53.
[44] J.A. Duffy and R.J. Grant, Phys. Chem. Glasscs 16 (1975) 124.
[45] H. Rawson, Inorganic glass-formingsystems (Academic Press, London, 1967) p. 112.
[46] P.J. Bray and J.G. O'Keefe, Phys. Chem. Glasses, 4 (1963) 37.
[47] J.A. Topping and J.O. Isard, Phys. Chem. Glasses 12 (1971) 145.
[48] J.O. Isard, J. Soc. Glass. Technol. 43 (1959) I13T.
[49] D.E. Day and G.E. Rindone, J. Am. Ceram. Soc. 45 (1962) 579.
[50] E.D. Lacy, Phys. Chem. Glasses 44 (1963) 234.
[51] R. Terai, Phys. Chem. Glasses 10 (1969) 146.
[52] B.E. Yoldas, Phys. Chem. Glasses 12 (1971) 28.
[53] D.D. Perrin, Dissociation constants of inorganic acids and bases in aqueous solution"
(Butterworths, London, 1969).
[54] J.A. Duffy and M.D. Ingram, J. Inorg. Nucl. Chem., to be published.
[55] G. Eisenman, ed., The glass electrode (Interscience, New York, 1966).
[56 ] G. Eisenman, ed., Glass electrodes for hydrogen and other cations (Arnold, London, 1967).
[57] F.G.K. Baucke, J. Non-Crystalline Solids 14 (1974) 13.
[58] A. Wikby, Phys. Chem. Glasses 15 (1974) 37.
[59] J.O. Isard, ref. 56, cha_pter 3.
[60] G.A. Perley, Anal. Chem. 21 (1949) 391.
[61] F.G.K. Baucke, in: Proceedings of 9th University Conference (Cleveland) on Ceramic
Science, ed. A.R. Cooper and A.H. Heuer (1972), to be published.
[62] F.G.K. Baucke, personal communication.
[63] H. Bach, Z. Angew. Phys. 28 (1970) 239;
[64] H. Bach and F.G.K. Baucke, Phys. Chem. Glasses 15 (1975) 123.
[65] H. Lux, Z. Electrochem. 45 (1939) 303.
[66] H. Flood and T. F~Srland, Acta Chem. Scand. 1 (1947) 592.
[67] A. Paul and R.W. Douglas, Phys. Chem. Glasses 8 (1967) 151.
[68] A.P. Herring, Phys. Chem. Glasses 7 (1966) 209.
[69] S. Holmquist, J. Am. Ceram. Soc. 56 (1973) 232.
[70] E.A. Guggenheim, J. Phys. Chem. 33 (1929) 842.
[71] R.W. Douglas, P. Nath and A. Paul, Phys. Chem. Glasses 6 (1965) 216; 7 (1966) 213.
[72] S.M. Budd, Phys. Chem. Glasses 7 (1966) 210.
[73] F.L. Harding, J. Am. Ceram. Soc. 55 (1972) 368; 56 (1973) 233.
[74] C.K. J~rgensen, Structure and Bonding, Vol. 1 (Springer-Verlag, Berlin, 1966) p. 234.
410
[75]
J.A. Duffy, M.D. Ingram
/An interpretation of glass chemistrv
P.G. Zambonin and J. Jordan, J. Am. Chem. Sot. 91 (1969) 2225.
[76] T. Moeller, Inorganic chemistry (Wiley, New York, 1952) p. 328.
[77] A.J. Easteal and D.J. Udy, J. Mater. Sci. 8 (1973) (17).
[78] R. Reisfeld and L. Boehm, J. Non-Crystalline Solids, 17 (1975) 209.